Re: [PATCH 3/4] ARM: change vmalloc_min to vmalloc_start

[Nicolas Pitre]

On Tue, 18 May 2021, Nicolas Pitre wrote:

> On Wed, 19 May 2021, Linus Walleij wrote:
> 
> > On Tue, May 18, 2021 at 2:15 PM Russell King (Oracle)
> > <rmk+kernel@xxxxxxxxxxxxxxx> wrote:
> > 
> > > Change the current vmalloc_min, which is supposed to be the lowest
> > > address of vmalloc space including the VMALLOC_OFFSET, to vmalloc_start
> > > which does not include VMALLOC_OFFSET.
> > >
> > > Signed-off-by: Russell King (Oracle) <rmk+kernel@xxxxxxxxxxxxxxx>
> > 
> > Reviewed-by: Linus Walleij <linus.walleij@xxxxxxxxxx>
> > 
> > > +static unsigned long __initdata vmalloc_start = VMALLOC_END - (240 << 20);
> > 
> > When I first read this it took me some time to figure out what was
> > going on here, so if you have time, please fold in a comment
> > with some explanation of that (240 << 20) thing, in some blog
> > post I described it as "an interesting way to write 0x0f000000"
> > but I suppose commit 0536bdf33faf chose this way for a
> > specific reason? (Paging Nico if he can explain it.)
> 
> That's an alternative (and deprecated) way to write MB(240).

And it seems that MB() isn't globally defined either. Oh well.


Nicolas